On Thu, Sep 26, 2013 at 2:34 AM, brian carroll <[email protected]> wrote: > coderman wrote: > >> you're wrong. > > perfect. thank you
brian: you're a verbose individual. but you respond usefully *grin* > i think i grasp a fundamental concept of crypto > that relates size of message (message length) > with design of algorithmic structure needed to > successfully embed or hide the message else > hidden order may be easily visible/discovered it is interesting how these fundamentals change across public key systems, and the ideal one time pad. symmetric ciphers are a particular beast... (and combined authentication and encryption modes even more particular ;) > i still contend this is different for set theory and > models of noise ... > > in that 'keys' could function differently in bit set > approach though perhaps rekeying is universal > as a security principle yet potentially flawed if > it could reveal a particular structure leading > to its compromise... in a poor implementation or protocol, re-keying can provide an opportunity for cipher suite downgrade or other privacy destroying attacks. effective frequent re-keying requires the other INFOSEC/OPSEC dependencies be met! > whereas reusing an 'infinity > key' (regenerating keys or using same key in > new instantiations, accessing different arbitrary > structure as keychain multitool) may function > in a different context than existing approaches, note that for all intents and purposes, you should use a fresh, absolutely random key for each re-keying. key "stretching" or derivation methods suffer the same types of vulnerabilities over large enough output that the original cipher does. instead of spending your time trying to securely "stretch" a few keys, just generate a large number of perfectly random keys instead!
